studies of inner-city women in London who had lost a parent in childhood, Brown and Harris (1989) demonstrated that those who did not have a close confiding relationship in adulthood and had unusually strong stressors developed depression. Those who had lost a parent in childhood but had good relationships and no such stressors did not develop depression. The authors also showed that the loss of care following the death of the parent was more important than the parental loss itself. Further studies established that lack of care in childhood following loss of a mother is associated with mood disturbance in adulthood. This influence is mediated by premarital pregnancy and marital dysfunction. Inadequate care increases the risk of early premarital pregnancy, which in turn increases the risk of marriage to an undependable partner. Marriage to such a partner, in addition to being low in intimacy, increases both the risk of serious life events like trouble with the law, threats of eviction, and poverty, and the risk of depression.
Recent evidence in adolescent females indicates that the accumulation of stresses at a time of transition may place them at greater risk for developing depressive symptoms (Petersen et al., 1992). In trying to understand why the rates of disorder are so high in adolescent women, Petersen and colleagues have suggested that the combination of going through puberty and having a transition from one form of school to another (from primary to secondary school), along with a breakup of the family, form a constellation of risk factors that leads to increased rates of depression in this group. Clearly, the burdens of childrearing, when not adequately supported, predispose to depression (Weissman, Leif, and Bruce, 1987b). Downward social mobility and the lack of access to opportunities also can play a role in depression (Dohrenwend et al., 1992).
Although in general, the foregoing risk factors are well established, detailed prospective studies of the influence of various risk factors over time and the mitigating effects of other variables, for the most part, have not been conducted in large samples. Two notable exceptions are the Epidemiologic Catchment Area (ECA) study (Anthony and Petronis, 1991) and the Alameda County study (Kaplan, Roberts, Camacho, and Coyne, 1987). Neither examined mood disorder in relatives, genetic influences, or the effect of severe trauma. They did, however, examine a variety of other influences. These studies excluded individuals who